International Journal of Hydrogen Energy 27 (2002) 507–526 www.elsevier.com/locate/ijhydene An assessment of alkaline fuel cell technology
G.F. McLean ∗, T. Niet, S. Prince-Richard, N. Djilali
University of Victoria, POB 3055, STN CSC Victoria, BC Canada V8W 3P6
Abstract This paper provides a review of the state of the art of alkaline fuel cell (AFC) technology based on publications during the past twenty-ÿve years. Although popular in the 1970s and 1980s, the AFC has fallen out of favour with the technical community in the light of the rapid development of Proton Exchange Membrane Fuel Cells (PEMFCs). AFCs have been shown to provide high power densities and achieve long lifetimes in certain applications, and appear to compete favourably with ambient air PEM fuel cells. In this report we examine the overall technology of AFCs, and review published claims about power density and lifetime performance. Issues surrounding the sensitivity of the AFC to CO2 in the oxidant stream are reviewed and potential solutions discussed. A rough cost comparison between ambient air AFCs and PEMFCs is presented. Overall, it appears the Alkaline Fuel Cell continues to have potential to succeed in certain market niche applications, but tends to lack the R& D support required to reÿne the technology into successful market o 1. Introduction published information on AFCs to provide a uniÿed view of the technology. A re-examination of the economics of AFC The alkaline fuel cell (AFC) was the ÿrst fuel cell tech- technology is also presented. The issues generally assumed nology to be put into practical service and make the gen- to have caused the demise of interest in AFCs, namely low eration of electricity from hydrogen feasible. Starting with power density and electrolyte poisoning are addressed in applications in space the alkaline cell provided high-energy detail to provide as complete a picture as possible, based conversion e=ciency with no moving parts and high reli- primarily on published and publicly available information. ability. AFCs were used as the basis for the ÿrst experi- The PEM fuel cell has recently emerged as the technol- ments with vehicular applications of fuel cells, starting with ogy of choice for low temperature, moderate power applica- a farm tractor in the late 1950s equipped with an Allis tions and has largely displaced the AFC in this application. Chalmers AFC (Kordesch and Simader, 1996). This was Because of this, we have provided a comparison between followed by the now famous Austin A40 operated by Karl alkaline and PEM technology wherever possible. In partic- Kordesch in the early 1970s [1] and continuing today with ular, a detailed cost comparison between PEM and AFCs is the commercialization activities of the ZEVCO company included. [2,3]. However, despite its early success and leadership role The public domain literature has been reviewed includ- in fuel cell technology, AFCs have fallen out of favour ing the most recently published results on alkaline elec- with the research community and have been eclipsed by the trode materials and manufacture as well as older publications rapid development of the Proton Exchange Membrane fuel describing the state of the art around 1980. Earlier publica- cell (PEMFC) as the technology of choice for vehicular tions, which largely describe the now defunct space appli- applications. cations of AFC technology, have not been reviewed. The This paper provides a critical overview of the state of overall purpose of this review has been to establish a techni- the art of AFC technology and attempts to synthesize the cal opinion about the viability of AFCs and to identify key areas for research. The report is structured as follows. In Section 2 we pro- ∗ Corresponding author. Fax: +250-721-6323. vide a general orientation to AFC technology and review E-mail address: [email protected] (G.F. McLean). the nature of the published research and recent corporate 0360-3199/02/$ 20.00 ? 2002 Published by Elsevier Science Ltd on behalf of the International Association for Hydrogen Energy. PII: S 0360-3199(01)00181-1 508 G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 et al. [5] and De Geeter [4], the active layer consists of an organic mixture (carbon black, catalyst and PTFE) which is ground, and then rolled at room temperature to cross link the powder to form a self supporting sheet. The hydrophobic layer, which prevents the electrolyte from leaking into the reactant gas Low channels and ensures di the presence of CO2, carbonates form and precipitate; − 2− CO2 + 2OH → (CO3) +H2O: The carbonates can lead to potential blockage of the elec- trolyte pathways and=or electrode pores. This issue is dis- cussed in detail in Section 3.2.1. The inherently faster kinetics of the oxygen reduction re- action in an alkaline cell allows the use of non-noble metal Fig. 1. Alkaline fuel cell composition. electrocatalysts. It is useful to compare the eletrochemical performance of AFCs and PEMFCs in terms of the relation- ship between cell potential, E, and current density, i. When activities. In Section 3 we discuss the major technical issues mass transport limitations are negligible (low to intermedi- confronting AFCs, including the reported power densities, ate current density), E and i are approximately related by poisoning issues, lifetime, duty cycles and systems consid- Blomen and Mugerwa [6] erations. This section also provides a hint at some new AFC technologies that may be of interest. Section 4 provides a de- E = E0 − ÿ log i − Ri tailed cost analysis and includes a comparison to published PEM cost projections. In Section 5 we provide conclusions with and state our general technical position. E0 = Er + ÿ log i0; 2. Alkaline fuel cell background and development status where, Er is the reversible thermodynamic potential, ÿ and i0 are the Tafel slope and the exchange current density for 2.1. Principle of operation the oxygen reaction, and R is the di The main contribution to cell resistance is due to the and mobile fuel cell applications since the mid 1980s. Ma- ionic=protonic resistivity of the electrolyte. Again AFCs jor European projects conducted by Siemens, Hoechts and appear to have lower electrolyte resistivities (0.05 vs. DLR were all cancelled prior to 1996 [21]. North American 0:08 P=cm2 for PEMFC). It should be pointed out that development of AFCs for space applications is continued new generation ultra thin acidic polymer membranes [8] by United Technologies=International fuel Cells. However, achieve low resistance. Nonetheless AFCs have an intrin- this work appears to be limited to providing fuel cells to the sic advantage over PEMFC on both cathode kinetics and space shuttle program and appears to have no aspirations for ohmic polarization. A puzzling aspect of all published AFC entering other markets [22]. data is that polarization curves are invariably presented for The remaining developers of AFC technology are almost maximum current densities of about 400 mA=cm2, with exclusively related to Zetek Corporation. Zetek is the par- no indication that mass transport limitations have been ent organization of three companies involved in developing reached. A possible explanation is that, for cost reasons, products for transportation (Zevco plc), marine (ZeMar Ltd) the catalysts of choice are nickel alloys. Nickel is, how- and stationary power (ZeGen Ltd) applications. Recent de- ever, susceptible to oxidation, leading to high performance velopments from Zetek include the announcement of a new degradation over time. This problem would presumably be 5MW automated production line in Germany that will see exacerbated at higher current densities. Zetek manufacture more fuel cells than the combined pro- duction capacity of the rest of the world [23]. 2.2. Research activity level Astris Energi [24] recently announced a 4-kW prototype systems and o The foregoing discussion serves to point out the di=culty 400 mA=cm2 is discussed brieLy, but no voltage informa- encountered in trying to assess the power densities reported tion is provided. Kordesch et al. [12] discuss system perfor- by AFC researchers. Results are often incomplete, and very mance, but like De Geeter do not provide a complete descrip- few reported results discuss system performance. Nonethe- tion of the operating parameters used in the system. Even less, we can review the partial results that have been reported so, current densities of 250 mA=cm2 are reported. Similar and infer from them a reasonable picture of the power den- performance is apparently achieved with Platinum, Silver or sity achieved by AFCs. a low cost Spinel catalyst. Weight and volume information is provided for the original Kordesch Austin fuel cell vehi- cle [10], but these data are now over 30 years old. Zevco’s 3.1.1. Space applications more recent designs easily supersede the original Kordesch The AFC was initially used in space applications to pro- system. Summary results with an Elenco stack are reported duce electrical power for mission critical services. As such, in Vegas et al. [14]. The reported data indicate very low these fuel cells were designed to provide reliable power, current densities of only 50 mA=cm2. with low volume and weight, at virtually unconstrained The systems discussed in the preceding paragraph are cost. based on unipolar cell construction. Performance of a bipo- A matrix type alkaline H –O fuel cell is discussed by 2 2 lar plate AFC are reported in Tomantschger et al. [30] Matryonin et al. [26]. The cell is indicated as operating at ◦ where current density similar to De Geeter is presented, 100 C and pressures of 4–4:5×105 Pa. The presentation dis- i.e. 100 mA=cm2 at 0:85 V running on air. The cell voltage cusses the e Table 1 Summary of space application AFC performance Operating point Power Press. Temp. Source (W=cm2) (psig) (◦C) mV mA=cm2 950 140 0.133 29 98 [26] 950 220 0.209 58 98 [26] 950 310 0.2945 116 98 [26] 950 150 0.1425 58 65 [26] 950 280 0.266 58 96 [26] 950 440 0.418 58 130 [26] 600 3200 1.92 58 98 [26] 600 4200 2.52 58 121 [26] 800 6730 5.384 299 149 [27] 740 1000 0.74 29 80 [28] 900 320 0.288 29 80 [28] 900 1000 0.9 60 80 [28] Table 2 Summary of terrestrial application AFC performance Current density Power at Point 2 Power at Point 2 Press. Temp. (◦C) Source 0:7 V (mA=cm2)0:7V(W=cm2) (W=cm2) (psig) & gases mV mA=cm2 290 0.203 800 260 0.208 atm H2–air 75 [12] 450 0.315 800 280 0.224 atm H2–air 75 [12] N=AN=A 670 100 0.067 atm [4] 90 0.063 800 35 0.028 44 H2–air 40 [31] 108 0.076 800 102 0.082 44 H2–air 40 [31] 115 0.081 570 225 0.128 atm H2–air 40? [32] 125 0.088 700 125 0.088 atm H2–O2 40 [32] 88 0.062 700 88 0.062 atm H2–air 40 [32] N=AN=A 750 186 0.140 atm H2–O2 40 [32] 157 0.110 700 157 0.110 atm H2–air 40 [32] N=AN=A 850 100 0.085 atm H2–air 65 [30] N=AN=A 900 100 0.090 atm H2–O2 65 [30] 87 0.061 670 100 0.067 atm H2–air 70 [2] 40 0.028 N=AN=AN=A atm H2–air 60 [14] extent of work being undertaken to reÿne and optimize the bed electrodes is described. Operation of an AFC using a performance of AFC systems. Looded gas di Table 3 Summary of ambient air PEM fuel cell performance Current density 0:7 V Power at 0:7 V Current density 0:6 V Power at 0:6 V Source (mA=cm2)(W=cm2) (mA=cm2)(W=cm2) 200 0.140 425 0.255 [50] 250 0.175 500 0.300 [48] 125 0.088 450 0.270 [49] Comparing the results presented in Tables 2 and 3, it is ap- A number of papers present a point blank dismissal of parent that available alkaline and PEM technologies achieve this problem, as illustrated by the following quote: “it is roughly equivalent current densities when operated on am- often reported that the AFC ... must be fed with pure oxygen bient air oxidant streams. This means that in applications because it is poisoned by CO2 in the atmosphere ... None where ambient air alkaline technology is proposed (as in of these myths can be substantiated” [4,15]. However, these Zevco’s planned hybrid vehicle system) there is no reason papers present no data to substantiate their claim. to think that the alkaline technology will be easily displaced Al Saleh et al. [37] showed that concentrations of up to ◦ by a better, more e=cient PEM system. 1% CO2 in the oxidant stream of Ag=PTFE electrode at 72 C did not signiÿcantly a Michael [2] reported that at 670 mV with 50 ppm CO2 With a regeneration period where the cells are run at a in the air stream over 6000 h the power output was re- higher current density performed at 7000 and 15; 000 h, the duced from 70 to 50 mW=cm2 (approximately a 30% re- lifetime of a cell was doubled to 20; 000 h [51]. No duction) for a 500 W stack. The paper stated that this was reference is given to substantiate this claim. Pratt and a non-continuous test but did not provide information on Whitney developed a system, which incorporated special electrolyte replenishment or replacement. regeneration cells into a regular fuel cell stack, regenerating A test with intermittent operation was also performed by the electrolyte continuously for the stack running on air. Zevco [2]. They found that the decrease in performance over They found that the loss of e=ciency was less than 1% time was greater with intermittent operation than with con- from the incorporation of these cells and that the cells could tinuous operation. However, the draining of the electrolyte run with 3000–4000 ppm carbon dioxide without a serious when the cells were shutdown seemed to prevent a large part e CO2 poisoning e CO2 adversely a CO2 in the oxidant stream has a distinct e The majority of published descriptions of AFC systems are based on the early work of Kordesch, followed by de- 3.3.1. Electrolyte circulation scriptions of Elenco [16,58] and then ZEVCO [4,12–14,30] The liquid electrolyte is circulated, allowing the possibil- systems. Complete lab scale systems are described in Kha- ity of removing product water and heat from the cell and lidi et al. [59] and Ergul [60]. These do not provide descrip- also allowing the possibility of removing carbonates from tions of practical fuel cell stacks, but do provide alternative the electrolyte to maintain cell performance. The circula- descriptions of means of electrolyte circulation, heat and tion of the electrolyte within the alkaline cell is analogous water management. Some discussion of stationary systems to the circulation of cooling within PEM cells with roughly 516 G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 equivalent complexity for both. The major di Table 4 Summary of reported Alkaline Fuel Cell lifetimes Date Hours Current density (mA=cm2) System information Source 1986 3500 100 Electrodes [30] 1987 5000 100 Elenco module [16] 1990 ¿ 2000 Unknown Siemens BZA4 [58] 1991 424& 600 100 Electrode tests [65] 1994 3500 100 Electrode tests [66] 1994 200 160 mV Overpotential Electrode tests [37,38] 1996 15,000 Unknown Not stated [48] 1996 15,000 Unknown Orbiter fuel cell [59] 1998 1000 Varied Elenco module [14] 1999 ¿ 5000 100 Zevco module [4] 1999 4000 Unknown From JPL [13] 1999 11,000 100 Anode electrode [32] 2000 6000 100 Zevco module [2] 2000 6000 100 Zevco module [2] performance. The magnitude of the improvement was not buildup, which could change the performance characteris- quantiÿed. tics of the test. Strasser [58] tested four Siemens BZA4 modules and found that each module showed a similar performance drop 3.4.3. Summary of AFC lifetimes over the course of the test. A drop of approximately 50 mV The lifetime of an AFC can, in general, be well over was observed over the course of this 700 h test performed at 5000 h for inexpensive terrestrial AFCs and has been shown ◦ 80 C with pure hydrogen and oxygen at 2.3 and 2:1 bars, re- to be signiÿcantly over 10; 000 h for space application AFCs. spectively. Tests of several thousand hours are brieLy men- It would not be unreasonable to assume, given a signiÿcant tioned but no details are provided. development e Table 5 Summary of low power ambient air fuel cell prices [71–73] Company (Fuel Cell Product) Nominal Power Type of Fuel Cell Price (US$) Astris (LC200-16) 240 W AFC 2400 H-Power (PowerPEM-PS250) 250 W PEMFC 5700 DAIS-Analytic (DAC-200) 200 W PEMFC 8500 these cost estimates to equivalent PEM costs. The reader is mention a 5–10 times higher production cost for small-scale cautioned that while the cost comparisons included in this production. section are based on the best available information at the Other projected general estimates for AFC material or time of writing, we have been faced with extrapolating costs stack costs range from US$ 80=kW to US$ 265=kW (ÿgures in some cases, estimating costs in others and comparing cost adjusted to 2000 US$) [18,30,13]. information provided for di Table 6 Materials and manufacturing processes for AFC stacks [4,5,21] Component Materials Manufacturing Processes Electrodes Anode PTFE powder Mechanical process involving graphite powder grinding, dispersion, ÿltering, catalyst: rolling and drying (Pt or Pd 0.12–0:5mg=cm2) Ni–Al, Ag Cathode PTFE powder Mechanical process involving graphite powder grinding, dispersion, ÿltering, catalyst: Pt rolling and drying White layer (for both anode and cathode) PTFE powder Pre-forming and rolling Module Current collectors Nickel mesh Pressed to black and white layers (as above) Plastic frames ABS plastic Injection molding and manual assembly with electrodes Spacers Unknown Stack assembly Plastic frames are friction-welded to module casing for sealing Table 7 Materials and manufacturing processes for PEMFC stacks [6,51] Component Materials Manufacturing Processes MEA Membrane Polymer matrix with attached Complex chemical process sulfonic acid groups E.g. Naÿon, BAM 3G, etc. Electrode substrate Carbon paper, PTFE Attached to membrane through hot pressing Catalyst Pt (0.4–4 mg=cm2) Deposited between the electrode substrate and the membrane Other Stack Components Flow ÿeld plates (including cooling plates) Graphite, stainless steel, Machined out of bulk material, carbon polymers, etc. stamped, injection molded Non-repeating components O< the shelf components Simple machining 4.2.2. PEMFC stack materials 4.2.3. AFC systemcosts Table 7 gives a summary of the PEMFC stack materials Table 8 lists the projected costs of a Zevco stack module. and manufacturing processes. This table is based on the data provided by Zevco in recent A number of potential improvements are foreseen for conferences and public presentations but has not been pub- both manufacturing and materials in PEMFC stacks. These lished in a citable reference. include the reduction of the catalyst loading down to No information on the assembly or manufacturing costs 0:04 mg=cm2 through improved deposition techniques, dif- has been speciÿcally stated for AFCs. However, it is reason- ferent nanostructure catalyst supports, the use of carbon able to assume that the manufacturing costs are included in composite materials and stamped metal sheets for the Low the component costs. The cost of ÿnal assembly, especially ÿeld plates and the reduction of the MEA thickness [68]. for larger volume manufacturing, is assumed to be minimal. 520 G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 Table 8 technological option there may be short-term cost advan- Costs of AFC stack components [8]∗ tages for other technologies. In particular, we are interested to determine if AFCs possess any inherent cost advantage Component Current Projected (US$=kW) (US$=kW) in small volume production that is more indicative of early fuel cell markets. Total stack costs 1750 205 No study presenting cost estimates of AFC stacks at very ∗Note: Converted from ECU=kW to US$=kW on a 1:0.925 basis. high volumes has been found. To estimate low volume production costs we have used the lowest power density ZEVCO Mark II costs described in the previous section. 4.2.4. PEMFC stack costs Conversely, our high volume cost estimate is derived on the PEMFC stack costs have been reported in a number of highest performance projections provided by ZEVCO. papers and reports, with current stack cost estimates rang- There are many sources providing high volume mass pro- ing from $500=kW [69] to $5000=kW [70]. Optimistic cost duction cost estimates of PEMFC systems. All these results projections for a 70 kW stack, for a typical automotive pro- are within the US$ 20–50=kW range. Directed Technolo- duction volume of 50,000 units per year, produce a lower gies completed one particularly thorough report for the Ford bound cost estimate of $20=kW [67]. Motor Company [67]. Most cost estimates for PEMFC sys- Table 9 summarizes the two extreme cost estimates avail- tem components at high volumes, used in the present anal- able in the literature. The data of Ekdunge and Raberg [70] ysis, are taken from this report. The Directed Technologies is summarized in the second column and covers material report provided cost estimates for PEM fuel cell systems in costs only for small-scale laboratory production of a 75 kW the 30–90 kW range. These cost estimates have been ex- unit using “conventional” materials. The data of James et al. trapolated to estimate the costs in the 7 kW range, details of [67], summarized in column 3, is on the other hand an es- this extrapolation are included in Appendix A. timate that includes material, manufacturing and assembly The estimated stack costs for both alkaline and PEM fuel costs for large-scale production of 30–90 kW stacks using cell technology are compared to each other at di Table 9 Costs of PEMFC components [67,70] Component 1998 PEMFC 500,000 unit per year % (500,000 units materials (US$=kW) production (US$=kW) per year) Membrane 120 0.40 2% Catalyst 243 8.20 41% Gas di Table 10 E AFC stack PEMFC stack $ PEMFC (US$=kW) (US$=kW) $ AFC Small batch fabrication 1750 2000–5000 1.2–2.9 [70,74] Small-scale manufacturing 205 500–1500∗ 2.5–7 (100s, 1000s?) [69,75] Improved AFC performance High volume production 155 20 [67] (50-kW unit) 0.13–0.4 (Unknown volume for AFC) 60 extrapolated from [67] (7-kW unit) (500,000 units=yr for PEMFC) ∗$1500=kW includes ancillaries. Table 11 circulation and nitrogen purging. PEM balance of plant re- Ambient air PEMFC cost estimate table quirements di the signiÿcant component of CO2 scrubbing because the 4.5. Balance of plant reactor vessel containing the soda lime is composed of a passive container operating without any high pressures or Balance of plant components that need to be considered temperatures. The cost of soda lime used over the lifetime for AFCs include the air blower, CO2 scrubber, electrolyte of the cell is provided in Section 4.6.1. 522 G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 Table 12 Table 13 AFC peripheral costs PEMFC peripheral costs Cost (US$) % Cost % (US$) Air blower 14 5.5% CO2 scrubber 14 5.5% Air compression subsystem 330 41.4% H2 recirculation ejector 22 8.6% (Compr.=Expander=Motor Unit—CMEU) Electrolyte recirculation 100 39.2% Air humidiÿer subsystem 65 8.1% Nitrogen purge 15 5.9% H2 recirculation ejector 22 2.8% Electronic engine control (EEC) 50 19.6% Radiator subsystem 92 11.5% Piping, valving, misc. 40 15.7% DI ÿlter 14 1.8% Electronic engine control (EEC) 220 27.5% Total periph. sys. cost 255 100% Piping, valving, misc. 55 6.9% (incl. mark-up and cost contingency) Total periph. sys. cost 798 100% (incl. mark-up and cost contingency) 4.5.1.3. Electrolyte recirculation loop. The main compo- nents of the alkaline electrolyte loop are the heater, a 50-W electrolyte pump, a small heat exchanger and a ventilator. Table 14 Estimated peripheral component costs for ambient air PEM systems The cost of this subsystem are assumed to be roughly equiv- alent to the cost of the coolant loop in conventional PEM Cost % technology, which we estimate to be US$ 100 in mass pro- (US$) duction [67]. Unlike PEM fuel cells, the electrolyte in AFCs requires maintenance and incurs an operational cost over the Air blower 14 5.5% Air humidiÿer subsystem 65 25.8% lifetime of the cell. This is discussed in Section 3.3.1. H2 recirculation ejector 22 8.6% Radiator subsystem 50 19.5% 4.5.1.4. Water management. Water management is rela- DI ÿlter 14 5.5% tively straightforward in the Zevco AFC system and incurs Electronic engine control (EEC) 50 19.5% a minimal cost. The associated components are mainly a Piping, valving, misc. 40 15.6% small water tank and a water condenser. Total periph. sys. cost 255 100% 4.5.1.5. Nitrogen purge. The use of a Nitrogen purge to (incl. mark-up and cost contingency) remove reactant gases from AFCs is shown on most system diagrams from the ZEVCO system [4,15]. However, no de- tails of the nitrogen purge system are available, and some tem. The air compressor is replaced with a simpler blower designs claim not to require this component. unit and the Electronic Engine Control is vastly reduced in complexity due to the simpler operation of the fuel cell as 4.5.2. Alkaline peripheral costs a battery charger. (Table 14) Based on the foregoing we are able to construct an esti- mate of the cost for balance of plant components required 4.6. Cost of consumables in AFC systems, as shown in Table 12. Most estimates are adapted from those of a PEMFC component with equivalent AFCs consume electrolyte and soda lime. In this section function. we estimate these costs. 4.5.3. Compressed PEMFC peripherals 4.6.1. Soda lime In a PEM fuel cell system the oxidant compression Soda lime is consumed in signiÿcant quantities in AFCs. system, electronic engine controls, radiator system and In fact it appears that the mass of soda lime used is approx- humidiÿer system contribute 89% of the cost of periph- imately equal to the mass of hydrogen used in normal cell eral components. An extrapolation to a 7 kW system operation. Therefore, regardless of the simple costs associ- of the peripheral costs given in the study by Directed ated with maintaining the soda lime scrubbing unit there is a Technologies=Ford (for systems in the range 30–90 kW) potentially large intangible cost associated with the regular [67] is presented in Table 13. maintenance required. Presently, scrubbing technology is able to make use of 4.5.4. Ambient air PEM peripherals only 7% of the limestone contained in the scrubber unit, but For an ambient air PEMFC system, the peripheral costs utilization up to 80% is achievable [2]. Using the rate of are somewhat di Table 15 tion volumes of 500,000 [67]. This means that extra compo- Soda lime cost estimate based on bulk cost of $ 0.2=kg, 5000 h nents required to complete the alkaline hybrid power system cell lifetime at 7 kW (namely batteries) must cost no more than about $670. % Utilization Consumption rate Lifetime mass Cost Of interest with this report is the opportunity available for CO2 scrubbing with novel technologies. The total cost of 7% 8 kW h=kg 4500 kg $900 CO2 scrubbing in the above system is $94 ($14 for the Can- 80% 92 kW h=kg 394 kg $80 ister and $80 for the Soda Lime). Therefore, if we consider the cost of the system without batteries or CO2 scrubbing we have a cost of $1334. This implies that, to be competi- Table 16 tive with a 50 kW PEMFC system, the cost of batteries and Cost of KOH electrolyte CO2 scrubbing must be less than about $750. Electrolyte Lifetime # Lifetime life (h) changes cost $US 5. Conclusions 300 17 153 500 10 90 In this report we have presented a review of AFC technol- 1000 5 45 ogy to assess its potential from both technical and economic 5000 1 9 perspectives. Research and development in AFC technology has become largely stagnant during the past decade although we can ÿnd no obvious technical or economic reasons for estimates for CO2 scrubbing over the 5000 h, lifetime of the the relative neglect it has received. system as shown in Table 15. AFCs can theoretically outperform PEMFCs and some of the earliest pressurized AFC systems showed current den- 4.6.2. KOH sities much higher than those achieved today with current A 7 kW AFC system requires 13 kg of 6–9 N KOH solu- PEM technology. Concerns about the low power density tion for the electrolyte, representing a mass of roughly 3 kg achieved by current AFC technology are misplaced, as the of KOH. Although KOH is considered to be a cheap bulk current AFC designs are directed at low power applications. material its cost must be factored into the overall system Ambient air operated AFCs produce current densities com- cost for an AFC. parable to ambient air operated PEMFCs. In small lab scale quantities KOH is available at Only a single design paradigm has been explored in com- US$9:00=kg, leading to an estimated cost of US $3.00=kg mercial AFC systems. There is considerable scope for im- in bulk. Based on this information and the frequency of provement of AFC technology through further research, in electrolyte replenishment over the lifetime of the cell the particular for the development of new architectures for AFC total electrolyte costs are estimated as shown in Table 16. operation. There is no strong IP position to prohibit the fur- ther development of AFCs, nor are there any material supply 4.7. Systemcost estimates issues to potentially impede AFC development. AFC tech- nology has the potential to yield major improvements for By combining the information presented in the previous modest R& D investments. sections upper and lower bounds for the cost of three com- Contamination of AFCs due to the presence of CO2 is an peting 7 kW fuel cell systems are produced, as shown in issue for sustained system operation. CO2 in the Cathode Table 17. Apart from being deduced from commercial data, air stream deÿnitely poisons the electrolyte and in turn can the alkaline estimates have a total cost range that is a factor cause some designs of electrodes to become clogged with of 6, compared to a range of 28 for the ambient air PEM es- carbonate. The use of high current draw from a cell to “elec- timate, reLecting our uncertainty here. Gulzow [21] quotes a trolyze” the carbonates should be investigated further. The ÿgure of $400–$500=kW for an AFC system using the tech- only practical solution to the CO2 problem currently em- nology of the time for high volume production. This number ployed is the use of Soda Lime for scrubbing CO2 from the falls in between the two ÿgures obtained for an AFC system. air-stream. This is cumbersome, comparatively costly (est. This analysis indicates AFC systems are cost competitive $US 94 per system) and has not been optimized as yet. De- with comparably sized PEMFC systems, at least for low velopment of new means of CO2 removal from the oxidant power. This advantage remains for all production volumes, stream for an AFC system would address many operational but is most signiÿcant at low and medium production vol- issues associated with AFC stacks. Contamination due to umes. impure hydrogen is another problem that may prohibit the However, it should be noted that the 7 kW alkaline system use of AFCs with reformed hydrogen streams, though this would be competing with a 50 kW PEMFC system. Directed contamination seems to be totally reversible. Technologies has estimated that the total system cost for a Current AFC systems have been demonstrated to easily 50 kW PEMFC system would be about $2100 for produc- meet the 5000 h lifetime required for traction applications. 524 G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 Table 17 Total system costs comparison Component Compressed PEMFC Ambient air PEMFC Ambient air alkaline Upper bound Lower bound Upper bound Lower bound Upper bound Lower bound Stack cost ($=kW) 1220 60 6100 180 643 155 Stack cost 8540 420 42,700 1260 10,942 1084 Balance of plant 798 798 256 256 255 255 Consumables N=AN=AN=AN=A 1053 89 Total 9338 1218 42,956 1516 12,250 1,428 Total per kW 1334 174 6136 217 1750 204 Table 18 PEMFC Power System Costs Cost summary for 7 kW PEMFC stack 3350 Total cost (US$) Cost per kW (US$=kW) 2850 Stack $419 $60 2350 System peripherals $798 $114 $ 1850 Total system cost $1217 $174 US 1350 850 350 However, electrolyte management issues in AFC’s imply a 010203040 50 60 70 80 90 100 degree of ongoing maintenance not necessary with PEMFC Fuel Cell System Net Power [kWe] technology. Periodic maintenance is also required for the FC Stack cost Total peripheral system cost Total cost Linear (FC Stack cost) Linear (Total peripheral system cost) existing soda lime CO2 scrubbing. Minimizing maintenance Linear (Total cost) in AFCs is an important topic for development. Fig. 2. PEMFC power system costs. Our analysis of costs shows that AFC systems for low power applications including hybrid vehicles are at least competitive with the cost of any equivalent system con- structed using PEMFC technology. The AFC system has a References low cost stack and low cost peripheral components. An am- bient air operated PEM system, while having low cost pe- [1] Kordesch K. Power Sources for Electric Vehicles. Modern ripheral components has prohibitively high stack costs. A Aspects of Electrochemistry, 10, Plenum Press, New York, high pressure PEM system, while enjoying low stack costs, 1975. p. 339–443. requires expensive peripheral components. Further improve- [2] Michael PD. An assessment of the prospects for fuel cell-powered cars. ETSU, United Kingdom, 2000. ments in AFC technology will only strengthen this compet- [3] Zevco Website, (www.zevco.co.uk), last accessed April 5, itive position. 2000. [4] De Geeter E, Mangan M, Spaegen S, Stinissen W, Vennekens G. Alkaline fuel cells for road traction. J Power Sources 1999;80:207–12. Appendix A: 7 kW PEMFC stack cost development [5] Kivisaari J, Lamminen J, Lampinen MJ, Viitanen M. Preparation and measurement of air electrodes for alkaline James et al. [67] develop stack and system costs for fuel cells. J Power Sources 1990;32:233–41. 30–90 kW PEMFC systems. The data given by James al- [6] Blomen LJMJ, Mugerwa MN. Fuel cell systems. New York: Plenum Press, 1993. lows for the extrapolation of these costs down to 7 kW. Fig- [7] Srinivasan S, Enayetullah MA, Somasundaram S, Swan DH, ure 2 illustrates this extrapolation. Manko D, Koch H, John Appleby A. Recent advancements It can be seen that the cost of the stack is much more sen- in solid polymer electrolyte fuel cell technology with sitive to scale than the cost of peripherals. The cost ÿgures low platinum loading electrodes. Proceedings of the 24th developed using this linear extrapolation for a 7 kW stack Intersociety Energy Conversion Engineering Conference, are given in Table 18. IEEE, 1989. p. 1623–9. G.F. McLean et al. / International Journal of Hydrogen Energy 27 (2002) 507–526 525 [8] Bahar B, Cavalca C, Cleghorn S, Kolde J. E [49] Sugawara Y, et al. High performance PEFC in dry condition at engines. M.A.Sc. Thesis, University of Victoria, Victoria, BC, atmospheric pressure. 1998 Fuel Cell Seminar, Palm Springs, 1996. 1998. [63] Ferguson J. The Future is Green for BWT, (dailynews.yahoo. [50] Koschany P. Small multi purpose fuel cell systems. Portable com/h/nm/20000313/tc/austria water 1.html), last accessed Fuel Cells Proceedings, Lucerne, Switzerland, 1999. 3=14=00. [51] Appleby AJ, Foulkes FR. Fuel cell handbook. Malabar, [64] Tomantschger K, Findlay R, Hanson M, Kordesch K, Florida: Krieger Publishing Company, 1993. Srinivasan S. Degradation modes of alkaline fuel cells and [52] Kinoshita K. Electrochemical oxygen technology. New York: their components. J Power Sources 1992;39:21–41. Wiley, 1992. [65] Lamminen J, Kivisaari J, Lampinen MJ, Viitanen M, [53] Ahuja V, Green RK. CO2 removal from air for alkaline Vuorisalo J. Preparation of air electrodes and long run tests. fuel cells operating with liquid hydrogen—heat exchanger J Electrochem Soc 1991;138:905–8. development. Int J Hydrogen Energy 1996;21:415–21. [66] Gulzow E, et al. Carbon dioxide tolerance of gas di